2 ms·

I'm pretty sure those numbers are for training, not inference. I've run it on _CPU_ and gotten ~1 token per second.